A two-layer affine optimization scheduling method for thermoelectric micro-energy grid considering source-load uncertainty

Abstract

The present invention relates to a two-layer affine optimization scheduling method for a thermoelectric micro-energy network considering source-load uncertainty. The method solves the conservative expansion problem of uncertainty variables caused by nonlinear operations during the optimization process by constructing a linear energy transmission model of a thermoelectric micro-energy network taking into account losses; and according to different optimization focuses of day-ahead scheduling, the influence of uncertain factors on the optimization scheduling is handled in a hierarchical manner. An interval linear programming model is constructed at the upper layer to solve the day-ahead uncertain unit combination problem. On this basis, an affine optimization model is constructed at the lower layer to solve the day-ahead uncertain scheduling optimization problem. The method can utilize the inherent characteristics of the affine algorithm to track the transmission trajectory of uncertain factors, quantitatively analyze the influence of each uncertain factor on the optimization scheduling, and improve the conservatism of the resulting solution.

Description

Technical Field

[0001] The invention relates to a double-layer affine optimization scheduling method for a thermoelectric micro-energy network considering source-load uncertainty. Background Art

[0002] The increasingly severe fossil energy depletion crisis and the urgent need to transform the traditional energy structure have made it a global consensus to develop an integrated energy system with higher energy utilization efficiency and multi-energy complementarity. However, as one of the integrated energy systems, the thermal power micro-energy network has a complex and changeable operating situation due to the random fluctuations of heterogeneous loads and the interactive coupling relationship of multiple types of energy. In addition, the increasing penetration rate of new energy to reduce carbon emission pressure ultimately makes the thermal power micro-energy network have a strong uncertainty operation characteristic, making its optimization scheduling problem complex and difficult to solve. Therefore, it is necessary to quantitatively analyze the impact of the random fluctuations of new energy output and load on the optimal scheduling of thermal power micro-energy networks, construct an uncertainty optimization scheduling model, realize the optimal scheduling of micro-energy networks considering multiple uncertainties, and improve the ability of micro-energy networks to cope with uncertainties.

[0003] At present, the uncertainty day-ahead optimization scheduling methods for thermal power micro-energy grids can be roughly divided into probabilistic methods and non-probabilistic methods. Among them, the probabilistic methods represented by random optimization and fuzzy methods require a precise probability distribution function or membership function of uncertain factors, and require a large amount of historical data to support the modeling of uncertain factors, which makes their application occasions quite limited; the non-probabilistic methods represented by robust optimization and interval optimization methods are easy to model, but the robust optimization method needs to ensure the feasibility of the solution under the worst scenario, and the interval optimization method cannot consider the correlation between uncertain variables, resulting in the obtained optimization scheduling scheme being too conservative. As an improved algorithm of the interval algorithm, the affine algorithm can consider the correlation between uncertain variables by sharing noise elements while retaining the advantages of the interval optimization method in easy modeling. However, the affine optimization method is still in the development stage and is mainly concentrated in a single power system. There is no effective day-ahead affine optimization scheduling method for thermal power micro-energy grids. Summary of the invention

[0004] The purpose of the present invention is to provide a double-layer affine optimization scheduling method for a thermoelectric micro-energy network considering source-load uncertainty, using an affine algorithm to track the inherent characteristics of the transmission trajectory of uncertain factors and quantitatively analyze the impact of various uncertain factors on the optimization scheduling.

[0005] To achieve the above object, the technical solution of the present invention is: a double-layer affine optimization scheduling method for a thermoelectric micro-energy network considering source-load uncertainty, comprising:

[0006] 1) According to the characteristics that the voltage of the power grid is near the per-unit value 1 during normal operation and the temperature difference between the supply and return water temperatures of each node in the heat network is small, the energy transmission process of the thermoelectric micro-energy network is linearized, and a linear energy transmission model of the thermoelectric micro-energy network is constructed, including a linear energy transmission model of the power system and a linear energy transmission model of the thermal system; using Taylor expansion, on the basis of the linear energy transmission model of the thermoelectric micro-energy network, the network loss of the thermoelectric micro-energy network operation is considered;

[0007] 2) Considering the impact of random fluctuations in renewable energy output and load on the optimal scheduling of thermal micro-energy networks, a two-layer affine optimization scheduling method for thermal micro-energy networks is proposed, with the upper layer being an interval linear programming model and the lower layer being an affine optimization model.

[0008] In one embodiment of the present invention, the above 1) is specifically implemented as follows:

[0009] 1.1) Construction of linear energy transfer model of power system

[0010] The power system energy transmission model is as follows:

[0011]

[0012] Where: and Inject active and reactive power into the nodes respectively; and are the active and reactive loads of the node respectively; and are respectively the active and reactive power transmitted by the line; V i t and V j t Represent the voltage amplitudes of nodes i and j respectively; represents the phase angle difference between nodes i and j; G ij and B ij are the real and imaginary parts of the node admittance matrix respectively; T represents the total number of scheduling time periods; N e Indicates the total number of power grid nodes;

[0013] Since the voltage of each node is close to the per-unit value 1 during the actual operation of the power system, the absolute value of the phase angle difference between adjacent nodes is within 10° and does not exceed 30°. and (V i t ) 2 ≈V i t , linearize equation (1) into equation (2), decoupling the grid state variables:

[0014]

[0015] The trigonometric function in equation (2) is expanded by the following first-order Taylor:

[0016]

[0017] Where: is the angle difference between nodes i and j at a certain operating point, is the voltage phase angle value of nodes i and j at this operating point;

[0018] make And use 1+ΔV i t Indicates V i t , ΔV i t is the node voltage deviation, then the power system energy transmission model can be transformed into formula (4):

[0019]

[0020] Where: g ij =-G ij , b ij =-B ij Respectively represent the line conductance and susceptance; y ij To accommodate the line to the ground;

[0021] Compared with formula (2), formula (4) can reflect the loss in the process of power grid energy transmission while ensuring the decoupling of state variables. The specific network loss constraint is shown in formula (5):

[0022]

[0023] 1.2) Construction of linear energy transfer model of thermal system

[0024] The energy transfer model of the thermal system is:

[0025]

[0026]

[0027]

[0028]

[0029] Where: Φ t is the thermal power vector, which represents the thermal power consumed or supplied by each node; C p is the specific heat capacity of water; A represents the node-pipeline association matrix; m t is the pipeline flow vector; and are the node heating temperature and outlet temperature vector respectively. When the node is a heat source node, the outlet temperature vector is the return water temperature of the node; and Respectively represent the flow temperature at the beginning and end of the pipeline; T a is the ambient temperature; λ is the total heat transfer coefficient per unit length of each pipe; L represents the length of the pipe; and They represent the pipe flow and its temperature before the node i respectively; and represents the pipe flow out of node i and the temperature of node i after the flow is mixed; B is the loop-pipe association matrix; K is the pipe resistance coefficient vector; f is the friction factor, whose size is determined by the Reynolds number R e Determine; D is the pipe diameter; ρ is the density of water; g represents the acceleration due to gravity;

[0030] Formula (6) defines the energy interaction relationship of the heating network to ensure the energy conservation of the heating network; Formula (7) reflects the temperature loss of the flow in the heating network pipeline; Formula (8) is the physical constraint that needs to be satisfied after the flow is mixed at a node; Formula (9) is the physical constraint that needs to be followed for the flow in the heating network pipeline to maintain circulation;

[0031] Based on formula (7) In actual operation, the term is far less than 1 and close to 0. The first-order Taylor expansion of the exponential term is performed to obtain the energy loss expression of the heat network pipeline flow (10):

[0032]

[0033] Where: is the energy loss caused by the flow passing through the pipeline between nodes i and j; They represent the energy of the flow at the beginning and end of the pipeline respectively;

[0034] It can be seen from formula (10) that after linearization, the heat network loss will be simplified to a constant term that is only related to the pipeline parameters itself; based on this, the heat network pipeline flow loss is incorporated into the load of each node, thus obtaining the heat network energy balance constraint shown in formula (11):

[0035]

[0036] Where: ΔΦ t Is The pipeline loss vector A l is the correlation matrix between the heat network node and the pipeline flow into the node;

[0037] In actual operation of the heating network, f=64 / R e , while R e =4|m t| / μρπD, μ is the kinematic viscosity of water, then equation (11) can be linearized into equation (12); so far, the energy transfer model of the thermal system is only a linear combination of the pipeline flow rate, realizing the linearization of the energy transfer model of the thermal system:

[0038]

[0039] Where: R f is the head loss coefficient vector.

[0040] In one embodiment of the present invention, in 2), based on the double-layer affine optimization scheduling method of the thermoelectric micro-energy network with the upper layer being the interval linear programming model and the lower layer being the affine optimization model, a double-layer affine optimization scheduling method of the thermoelectric micro-energy network is proposed for different needs of the thermoelectric micro-energy network day-ahead scheduling; the details are as follows:

[0041] 2.1) Day-ahead optimization objective function

[0042] While ensuring optimal scheduling and taking into account carbon emission requirements, the objective function constructed is shown in formula (13), which includes minimizing the unit capacity cost, carbon emission penalty cost, equipment operation and maintenance cost, energy abandonment cost, main grid power exchange cost and unit start-up and shutdown cost:

[0043]

[0044] in:

[0045]

[0046] Where: F fuel is the capacity cost of the CHP unit, F i steam is the capacity cost of the steam CHP unit, F i gas is the capacity cost of the gas-fired CHP unit, and are the electrical output and thermal output of the steam CHP unit, is the electrical output of the gas-fired CHP unit, N s and N g are the total number of steam CHP units and gas CHP units, is a 0-1 variable, indicating the working status of the gas-fired CHP unit; F co2 is the carbon emission penalty cost, γ represents the carbon emission intensity of the CHP unit; F om is the equipment operation and maintenance cost, N d is the total number of devices, F is the equipment output value; s is the energy abandonment cost, and are the values ​​of abandoned solar power, abandoned wind power and abandoned load respectively; F trade is the main network power interaction cost, and Respectively represent the purchase and sale of electricity to the main network; F UC is the unit start-up and shutdown cost, and are 0-1 variables, representing the startup and shutdown actions of the two types of CHP units, j∈{s,g}; c co2 , c sPv 、c sPw 、c sp , They represent the corresponding cost coefficients respectively; Δt is the duration of a single scheduling period;

[0047] 2.2) Upper level interval linear programming model

[0048] An upper interval linear programming model is constructed, and interval numbers are used to describe the random fluctuations of renewable energy output and load. An interval linear optimization method is proposed to solve the day-ahead unit combination problem of thermal power micro-energy grid. The details are as follows:

[0049] When the output and load of renewable energy are modeled as interval numbers according to the predicted fluctuation boundary, the continuous variables including unit output, pipeline flow, node voltage, and energy storage device output will be transformed into corresponding interval forms, so that the unit combination problem of the thermal power micro-energy network becomes an uncertainty interval optimization problem; the interval optimization problem is solved by transforming the original uncertainty optimization model containing interval variables into two deterministic sub-models, namely the upper limit sub-model and the lower limit sub-model, so as to obtain the upper limit of the cost under the worst conditions and the lower limit of the cost under the optimal conditions, thereby solving the interval optimization problem; based on the constructed linear energy transmission model of the thermal power micro-energy network, the constraints in the interval optimization model are all linear constraints; on this basis, considering that the objective function is a monotonic convex function, it can be known that The worst and best conditions that make the objective function take the upper and lower limit values ​​will appear on the fluctuation boundary of the new energy output and load; combined with the highly positive correlation between the objective function value and the unit output, it can be obtained that the worst condition is the maximum unit output, and the best condition is the minimum unit output, that is, the worst condition is that the new energy output changes to the lower limit of the interval and the load changes to the upper limit of the interval, and the best condition is that the new energy output changes to the upper limit of the interval and the load changes to the lower limit of the interval; the worst and best conditions can be used to transform the uncertain interval optimization model into a deterministic upper limit sub-model and a lower limit sub-model. By solving the corresponding sub-models, the uncertain unit combination problem can be solved, and a unit combination interval scheduling plan is provided for the construction of the lower affine optimization model;

[0050] 2.3) Lower layer affine optimization model

[0051] Based on the unit combination interval scheduling plan obtained by solving the upper interval linear programming model, the initial optimal operating point of each unit combination plan is further solved when the output of new energy and load fluctuations are not considered. The Taylor expansion mentioned in the linear energy transmission model of the thermal micro-energy network is performed at this operating point, and the linear energy transmission model of the thermal micro-energy network taking into account the loss is constructed. On this basis, the lower-level affine optimization model considering the uncertainty of new energy output and load fluctuations is constructed.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] (1) The linear energy transfer model simplifies the nonlinear operations in the day-ahead scheduling optimization process of the thermal power micro-energy grid and solves the conservative expansion problem caused by nonlinear operations of uncertain variables.

[0054] (2) By taking the loss into account in the linear energy transfer model, the affine algorithm is used to eliminate the adverse effects caused by the uncertainty sources that cannot be taken into account in interval optimization, thereby reducing the conservatism of the optimization results.

[0055] (3) A convex optimization affine model construction method suitable for fast solution by commercial solvers realizes the affine domain definition and convex processing of complex operating constraints of thermal power micro-energy grids, solving the problem of difficulty in constructing affine optimization models for thermal power micro-energy grids.

[0056] (4) Aiming at the different optimization requirements of the day-ahead dispatch of the thermal power micro-energy grid, the present invention establishes an interval linear programming model at the upper level to solve the uncertainty unit combination problem, and establishes an affine optimization model at the lower level to solve the uncertainty dispatch optimization problem, and hierarchically handles the impact of uncertain factors on day-ahead dispatch. The affine algorithm is used to track the impact trajectory of uncertain factors, characterize the impact of each uncertain factor on the optimized dispatch, and realize the location of sensitive points and weak points of system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0058] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0059] The present invention provides a double-layer affine optimization scheduling method for a thermoelectric micro-energy network considering source-load uncertainty, comprising:

[0060] 1) According to the characteristics that the voltage of the power grid is near the per-unit value 1 during normal operation and the temperature difference between the supply and return water temperatures of each node in the heat network is small, the energy transmission process of the thermoelectric micro-energy network is linearized, and a linear energy transmission model of the thermoelectric micro-energy network is constructed, including a linear energy transmission model of the power system and a linear energy transmission model of the thermal system; using Taylor expansion, on the basis of the linear energy transmission model of the thermoelectric micro-energy network, the network loss of the thermoelectric micro-energy network operation is considered;

[0061] 2) Considering the impact of random fluctuations in renewable energy output and load on the optimal scheduling of thermal micro-energy networks, a two-layer affine optimization scheduling method for thermal micro-energy networks is proposed, with the upper layer being an interval linear programming model and the lower layer being an affine optimization model.

[0062] The following is the specific implementation process of the present invention.

[0063] The flowchart of the double-layer affine optimization scheduling method for the thermoelectric micro-energy network considering source-load uncertainty proposed in the present invention is as follows: Figure 1 As shown, including:

[0064] 1) According to the characteristics that the voltage of the power grid is close to the per-unit value 1 during normal operation and the temperature difference between the supply and return water temperatures of each node in the thermal network is small, the energy transmission process of the thermal power micro-energy network is linearized, and a linear energy transmission model of the power system and the thermal system is constructed.

[0065] 2) Using Taylor expansion, on the basis of the linear energy transmission model of the thermoelectric micro-energy grid, the network loss of the thermoelectric micro-energy grid is considered.

[0066] 3) On the premise of ensuring completeness, a convex optimization affine model construction method for thermoelectric micro-energy network suitable for fast solution by commercial solvers is proposed.

[0067] 4) Considering the impact of new energy output and random fluctuations in load on the optimal scheduling of thermal power micro-energy networks, a two-layer affine optimization scheduling method for thermal power grid and micro-energy networks is proposed, with the upper layer being an interval linear programming model and the lower layer being an affine optimization model. This method realizes the application of affine optimization method in the day-ahead scheduling of thermal power micro-energy networks, overcoming the modeling difficulties of traditional probabilistic optimization methods and the conservative tendency of non-probabilistic optimization methods.

[0068] The specific technical solution is described as follows:

[0069] 1. Linear energy transfer model of thermoelectric micro-energy network

[0070] The energy transmission model of the thermoelectric micro-energy network has significant nonlinearity, which greatly increases the difficulty of solving the day-ahead scheduling optimization problem. Therefore, the present invention linearizes it and constructs a linearized energy transmission model of the power system and a linearized energy transmission model of the thermal system respectively.

[0071] 1.1 Power system linear energy transfer model

[0072] The power system energy transmission model is:

[0073]

[0074] Where: and Inject active and reactive power into the nodes respectively; and are the active and reactive loads of the node respectively; and are respectively the active and reactive power transmitted by the line; V i t and V j t Represent the voltage amplitudes of nodes i and j respectively; represents the phase angle difference between nodes i and j; G ij and B ij are the real and imaginary parts of the node admittance matrix respectively; T represents the total number of scheduling time periods; N e Indicates the total number of power grid nodes;

[0075] Considering that the voltage of each node is close to 1 per unit when the power system is actually running, and the absolute value of the phase angle difference between adjacent nodes is basically within 10° and does not exceed 30°, it can be assumed that and (V i t ) 2 ≈V i t , linearize equation (1) into equation (2), decoupling the grid state variables:

[0076]

[0077] Furthermore, the trigonometric function in equation (2) can be expanded by the following first-order Taylor expansion:

[0078]

[0079] Where: is the angle difference between nodes i and j at a certain operating point, is the voltage phase angle value of nodes i and j at this operating point;

[0080] make And use 1+ΔV i t Indicates V i t , ΔV i t is the node voltage deviation, then the power system energy transmission model can be transformed into formula (4):

[0081]

[0082] Where: g ij =-G ij 、b ij =-B ij Respectively represent the line conductance and susceptance; y ij To accommodate the line to the ground;

[0083] Compared with formula (2), formula (4) can reflect the loss in the process of power grid energy transmission while ensuring the decoupling of state variables. The specific network loss constraint is shown in formula (5):

[0084]

[0085] 1.2. Construction of linear energy transfer model of thermal system

[0086] The energy transfer model of the thermal system is:

[0087]

[0088]

[0089]

[0090]

[0091] Where: Φ t is the thermal power vector, which represents the thermal power consumed or supplied by each node; C p is the specific heat capacity of water; A represents the node-pipeline association matrix; m t is the pipeline flow vector; and are the node heating temperature and outlet temperature vector respectively. When the node is a heat source node, the outlet temperature vector is the return water temperature of the node; and Respectively represent the flow temperature at the beginning and end of the pipeline; T a is the ambient temperature; λ is the total heat transfer coefficient per unit length of each pipe; L represents the length of the pipe; and They represent the pipe flow and its temperature before the node i respectively; and represents the pipe flow out of node i and the temperature of node i after the flow is mixed; B is the loop-pipe association matrix; K is the pipeline resistance coefficient vector; f is the friction factor, whose size is determined by the Reynolds number R e Determine; D is the pipe diameter; ρ is the density of water; g represents the acceleration due to gravity;

[0092] Formula (6) defines the energy interaction relationship of the heating network and ensures the energy conservation of the heating network; Formula (7) reflects the temperature loss of the flow (mass flow rate) in the heating network pipeline; Formula (8) is the physical constraint that needs to be satisfied after the flow is mixed at a certain node; Formula (9) is the physical constraint that needs to be followed for the flow in the heating network pipeline to remain flowing.

[0093] Considering that the temperature difference between the nodes of the heat supply and return water networks in actual operation is basically 1°C and does not exceed 2°C, which is much smaller than the temperature drop caused by the supply heat load, the present invention assumes that the heating temperature and return water temperature of each node are equal to different constants, so as to achieve the decoupling of node temperature and pipeline flow in the energy transmission model. On this basis, the present invention is based on formula (7) In actual operation, the term is far less than 1 and close to 0. The first-order Taylor expansion of the exponential term is performed to obtain the energy loss expression of the heat network pipeline flow (10):

[0094]

[0095] Where: is the energy loss caused by the flow passing through the pipeline between nodes i and j; They represent the energy of the flow at the beginning and end of the pipeline respectively;

[0096] It can be seen from formula (10) that after reasonable linearization, the heat network loss will be simplified to a constant term that is only related to the pipeline's own parameters; based on this, the present invention adopts the concept of calculated load in power system analysis and incorporates the heat network pipeline flow loss into each node load, thereby obtaining the heat network energy balance constraint shown in formula (11):

[0097]

[0098] Where: ΔΦ t Is The pipeline loss vector A l is the correlation matrix between the heat network node and the pipeline flow into the node;

[0099] Finally, considering that in actual network operation, f = 64 / R e , while R e =4|m t | / μρπD, μ is the kinematic viscosity of water, then equation (11) can be linearized into equation (12); so far, the energy transfer model of the thermal system is only a linear combination of the pipeline flow rate, realizing the linearization of the energy transfer model of the thermal system:

[0100]

[0101] Where: R f is the head loss coefficient vector.

[0102] 2. Day-ahead double-layer affine optimization scheduling method for thermoelectric micro-energy grid

[0103] Based on the aforementioned linear energy transfer model, the present invention solves the conservative expansion problem caused by nonlinear operations in the optimization process of uncertain variables. On this basis, a two-layer affine optimization scheduling method for the day-ahead scheduling of the thermal power micro-energy network is proposed to meet the different needs of the day-ahead scheduling of the thermal power micro-energy network. The upper layer uses the interval linear programming method to solve the problem of uncertain unit combination, and the lower layer uses the affine optimization method to solve the problem of uncertain scheduling optimization.

[0104] 2.1. Day-ahead optimization objective function

[0105] While ensuring optimal scheduling and taking into account carbon emission requirements, the objective function constructed is shown in formula (13), which includes minimizing the unit capacity cost, carbon emission penalty cost, equipment operation and maintenance cost, energy abandonment cost, main grid power exchange cost and unit start-up and shutdown cost:

[0106]

[0107] in:

[0108]

[0109] Where: F fuel is the capacity cost of the combined heat and power (CHP) unit, F i steam is the capacity cost of the steam CHP unit, F i gas is the capacity cost of the gas-fired CHP unit, and are the electrical output and thermal output of the steam CHP unit, is the electrical output of the gas-fired CHP unit, N s and N g are the total number of steam CHP units and gas CHP units, is a 0-1 variable, indicating the working status of the gas-fired CHP unit; F co2 is the carbon emission penalty cost, γ represents the carbon emission intensity of the CHP unit; F om is the equipment operation and maintenance cost, N d is the total number of devices, F is the equipment output value; s is the energy abandonment cost, and are the values ​​of abandoned solar power, abandoned wind power and abandoned load respectively; F trade is the main network power interaction cost, and Respectively represent the purchase and sale of electricity to the main network; F UC is the unit start-up and shutdown cost, and are 0-1 variables, representing the startup and shutdown actions of the two types of CHP units, j∈{s,g}; c co2 , c sPv 、c sPw 、c sp , They represent the corresponding cost coefficients respectively; Δt is the duration of a single scheduling period;

[0110] 2.2 Upper Interval Linear Programming Model

[0111] Considering that the start and stop of the unit is a major decision in the day-ahead scheduling, it has a high cost and is less affected by the energy transmission loss. In addition, the start and stop decision variables are 0-1 variables. Unlike continuous decision variables such as unit output, the optimization space of the unit start and stop decision variables is small. Therefore, the present invention adopts a hierarchical optimization method to optimize the uncertainty unit combination problem of the unit start and stop variables. Without considering the loss, an upper interval linear programming model is constructed, and interval numbers are used to describe the random fluctuations of new energy output and load. An interval linear optimization method is proposed to solve the day-ahead unit combination problem of the thermal power micro-energy network, and provide a unit combination interval solution for constructing the lower-level day-ahead scheduling affine optimization model.

[0112] When the output and load of renewable energy are modeled as interval numbers according to the predicted fluctuation boundary, continuous variables such as unit output, pipeline flow, node voltage, and energy storage device output will all be transformed into corresponding interval forms, turning the combination problem of thermal power micro-energy network units into an uncertainty interval optimization problem. The solution to the interval optimization problem can be achieved by transforming the original uncertainty optimization model containing interval variables into two deterministic sub-models, namely the upper limit sub-model and the lower limit sub-model, so as to obtain the upper limit value of the cost under the worst conditions and the lower limit value of the cost under the best conditions, thereby solving the interval optimization problem. Based on the linear energy transfer model constructed above, the constraints in the interval optimization model are all linear constraints. On this basis, considering that the objective function is a monotonic convex function, it can be seen that the worst and best conditions for the objective function to take the upper and lower limit values ​​will appear on the fluctuation boundary of renewable energy output and load. Combined with the highly positive correlation between the objective function value and the unit output, the worst condition is the maximum unit output, and the best condition is the minimum unit output, that is, the worst condition is that the new energy output changes to the lower limit of the interval and the load changes of each node are the upper limit of the interval, and the best condition is that the new energy output changes to the upper limit of the interval and the load changes of each node are the lower limit of the interval. The worst and best conditions can be used to transform the uncertain interval optimization model into a deterministic upper limit sub-model and lower limit sub-model. By solving the corresponding sub-models, the uncertain unit combination problem can be solved, providing a unit combination interval scheduling solution for the construction of the lower affine optimization model.

[0113] 2.3. Lower layer affine optimization model

[0114] Based on the unit combination interval scheduling scheme obtained by solving the upper interval linear programming model, the initial optimal operating point of each unit combination scheme can be further solved without considering the output of new energy and load fluctuations. The Taylor expansion mentioned in the aforementioned linear energy transfer model is performed at this operating point, and a linear energy transfer model taking into account losses can be constructed. On this basis, the present invention will construct a lower-level affine optimization model that considers the uncertainty of new energy output and load fluctuations to solve the day-ahead scheduling optimization problem of thermal power micro-energy network.

[0115] 1) Affine Basic Theory

[0116] The definition of affine variables is shown in formula (15), which can track the source of uncertain factors through noise elements, use noise element coefficients to characterize the magnitude of the influence of uncertain factors, and reflect the correlation between each other in the form of sharing the same noise elements between uncertainty variables. The upper and lower limits of affine variables can be calculated by formula (16). Based on the linear energy transfer model taking into account losses, the present invention can solve the conservative expansion problem caused by the addition of affine noise elements in nonlinear operations in the affine algorithm during the optimization process.

[0117]

[0118]

[0119] Where: is the affine center value, which represents the determined part of the affine quantity; is the noise element, which indicates the fluctuation of the uncertainty factor i. is the corresponding noise element coefficient, indicating the fluctuation limit of the uncertainty factor; n is the total number of noise elements; is the upper limit value; x t is the lower limit value.

[0120] 2) Affine expression of uncertain factors

[0121] The present invention retains the advantage of easy modeling of the interval optimization method and realizes the definition of the affine expression of uncertain factors, that is, using the new energy output and load fluctuation boundaries required for interval optimization modeling, and obtaining the affine expression of uncertain factors as shown in formula (17) by inversely deducing formula (16).

[0122]

[0123] Where: and Represent the affine forms of active load, thermal load, photovoltaic output and wind power output respectively; and Indicates the corresponding central value, that is, the predicted value; and is the corresponding noise element coefficient, i.e., the fluctuation limit; ε represents the corresponding noise element; N h is the total number of heating network nodes.

[0124] 3) Definition of affine operation constraints of thermoelectric micro-energy network

[0125] 3.1) Inequality constraints

[0126] When the unit combination scheme is determined, the real domain inequality operation constraints of the thermal power micro-energy network can be divided into three categories: the value range constraints of a single continuous variable (such as state variables, unit output constraints, etc.), the combination constraints containing a single continuous variable and a 0-1 variable (such as the main network exchange power, energy storage equipment charging and discharging constraints) and the combination constraints of two continuous variables (unit climbing constraints). Under the premise of ensuring the completeness of the optimization, the present invention sequentially gives its affine domain definition method as shown in equations (18), (19) and (20). In addition, the affine inequality constraint definition method proposed in the present invention is not only applicable to the definition of thermal power micro-energy network operation constraints, but also to other inequality constraints with the same variable composition.

[0127]

[0128]

[0129]

[0130] Where: and are the upper and lower limits of the variable respectively; U is a 0-1 variable; is the ramp rate limit.

[0131] Since equations (18), (19) and (20) all contain absolute values, if not processed, the linear nature of the constraints will be destroyed, making the model difficult to solve. Therefore, the present invention uses equation (21) to perform equivalence on the absolute values.

[0132]

[0133] Where: is an auxiliary variable used for equivalent

[0134] Since the optimization goal of the model is to minimize the cost, There will not be a situation where both are non-zero, ensuring Instead of appearing So you can use The absolute values ​​in (18), (19), and (20) ensure the linear nature of the affine inequality constraints.

[0135] 3.2) Equality Constraints

[0136] When the unit combination is determined, there are three types of real-domain equation operation constraints for the thermal power micro-energy network: power balance constraints, energy storage equipment capacity state change constraints, and combined heat and power unit output coupling constraints. The first two types of equation constraints are based on the linear energy transfer model taking into account losses. They belong to equation constraints with the same noise element affine number. According to the judgment method of equal affine numbers with the same noise element, the central value of the affine numbers at both ends of the equation is equal to the noise element coefficient. The definition of the affine equation constraint can be achieved by replacing the variables in the original real-domain equation operation constraint with the affine center value xt 0 and the noise element coefficient xt i respectively. For the last type of operation constraint, that is, the combined heat and power unit output coupling constraint, since the uncertain factors (noise elements) in the electricity and heat networks are different from each other, the present invention defines it as an affine domain equation operation constraint as shown in formula (22), and ensures the completeness of the scheduling optimization by ensuring that it can still satisfy the thermal power output coupling relationship under the fluctuation limit value. In addition, the affine domain definition method of the power balance constraint, energy storage device capacity state change constraint and cogeneration unit output coupling constraint proposed in the present invention is also applicable to other integrated energy systems.

[0137]

[0138] Where: are the central values ​​of the electrical output and thermal output of the gas-fired CHP unit, respectively; is the noise element coefficient of the CHP unit's electrical and thermal output; R hp,i is the heat-to-electricity ratio of the gas-fired CHP unit.

[0139] 4) Affine objective function definition

[0140] When the unit combination is determined, the unit start-up and shutdown cost is a constant, which can be recorded as FUC,0. Based on the basic structure of affine variables as shown in formula (23), the present invention defines the original real number domain day-ahead optimization objective function as an affine domain objective function as shown in the formula. It can be seen from the formula that the present invention realizes the day-ahead scheduling optimization of the thermal power micro-energy network taking into account the uncertainty of source-load by dividing the optimization target into two parts: the central value cost and the fluctuation cost. Among them, the central value cost item refers to the scheduling and operation cost of the thermal power micro-energy network when the output and load of new energy are the basic situation (predicted value or affine central value); the fluctuation cost item refers to the correction scheduling cost of the thermal power micro-energy network when the output and load of new energy deviate from the predicted value and undergo random fluctuations; the value of w can be adjusted according to the conservative requirements of the dispatching management personnel on the fluctuation cost of the thermal power micro-energy network.

[0141]

[0142] Where: F fuel,0 、F co2,0 、F om,0 、F s,0 、F trade,0 is the center value cost; F fuel,e 、F co2,e 、F om,e 、F s,e 、F trade,e is the fluctuation cost; w is the optimization weight, and its value range is [0,1].

[0143] The central value cost in formula (23) can be constructed by replacing the variables in the original real domain objective function formula (13) with the corresponding central value xt 0; the fluctuation cost can be constructed by replacing the variables in formula (13) with the corresponding fluctuation limit To speed up the solution of the affine optimization model, the present invention introduces a second-order cone relaxation to process the objective function. By introducing the auxiliary variable S * t By replacing the quadratic term in the objective function, the original optimization problem is transformed into a second-order cone programming (SOCP) problem. The specific transformation method is shown in Equation (24). Since the optimization goal is to minimize the total operating cost of the thermal power micro-energy grid, that is, S* t Will continue to decrease, so in the process of optimization solution, it will continue to approach the direction of equality, and finally achieve the tightening of the second-order cone relaxation, thereby ensuring the feasibility and effectiveness of the obtained optimization scheduling solution.

[0144]

[0145] Where: f i t is the quadratic term of the replaced objective function, and c is the corresponding cost coefficient.

[0146] 3. Double-layer affine optimization method

[0147] In summary, the present invention proposes a two-layer affine optimization method for the day-ahead scheduling of a thermal power micro-energy network, which deals with the influence of uncertain factors on the optimal scheduling in a hierarchical manner according to different optimization focuses of the day-ahead scheduling. The upper layer constructs an interval linear programming model to solve the uncertainty unit combination problem, obtain the unit combination interval plan, and provide a specific unit start-stop plan and the initial optimal operating point under the plan for the construction of the lower-layer affine optimization model. On this basis, the lower layer establishes an affine optimization model, uses an affine algorithm to track the essential characteristics of the uncertainty source, analyzes the influence of various uncertain factors on the optimal scheduling, solves the uncertainty scheduling optimization problem, and takes into account the relationship between uncertainty variables in the form of shared noise elements, so that the proposed method has better economy and lower conservatism than the interval optimization method.

[0148] The above are preferred embodiments of the present invention. Any changes made according to the technical solution of the present invention, as long as the resulting functions do not exceed the scope of the technical solution of the present invention, belong to the protection scope of the present invention.

Claims

1. A double-layer affine optimization scheduling method for a thermoelectric micro-energy network considering source-load uncertainty, characterized by comprising: 1) According to the characteristics that the voltage of the power grid is near the per-unit value 1 during normal operation and the temperature difference between the supply and return water temperatures of each node in the heat network is small, the energy transmission process of the thermoelectric micro-energy network is linearized, and a linear energy transmission model of the thermoelectric micro-energy network is constructed, including a linear energy transmission model of the power system and a linear energy transmission model of the thermal system; using Taylor expansion, on the basis of the linear energy transmission model of the thermoelectric micro-energy network, the network loss of the thermoelectric micro-energy network operation is considered; 2) Considering the impact of new energy output and random load fluctuations on the optimal scheduling of thermal power micro-energy networks, a double-layer affine optimization scheduling method for thermal power micro-energy networks is proposed, with the upper layer being an interval linear programming model and the lower layer being an affine optimization model; The specific implementation of 1) is as follows: 1.1) Construction of linear energy transfer model of power system The power system energy transmission model is as follows: Where: and Inject active and reactive power into the nodes respectively; and are the active and reactive loads of the node respectively; and are respectively the active and reactive power transmitted by the line; V i t and Represent the voltage amplitudes of nodes i and j respectively; represents the phase angle difference between nodes i and j; G ij and B ij are the real and imaginary parts of the node admittance matrix respectively; T represents the total number of scheduling time periods; N e Indicates the total number of power grid nodes; Since the voltage of each node is close to the per-unit value 1 during the actual operation of the power system, the absolute value of the phase angle difference between adjacent nodes is within 10° and does not exceed 30°. and Linearize equation (1) into equation (2) to decouple the grid state variables: The trigonometric function in equation (2) is expanded by the following first-order Taylor: Where: is the angle difference between nodes i and j at a certain operating point, is the voltage phase angle value of nodes i and j at this operating point; make And use 1+ΔVi t Indicates V i t , ΔV i t is the node voltage deviation, then the power system energy transmission model can be transformed into formula (4): Where: g ij =-G ij 、b ij =-B ij Respectively represent the line conductance and susceptance; y ij To accommodate the line to the ground; Compared with formula (2), formula (4) can reflect the loss in the process of power grid energy transmission while ensuring the decoupling of state variables. The specific network loss constraint is shown in formula (5): 1.2) Construction of linear energy transfer model of thermal system The energy transfer model of the thermal system is: Where: Φ t is the thermal power vector, which represents the thermal power consumed or supplied by each node; C p is the specific heat capacity of water; A represents the node-pipeline association matrix; m t is the pipeline flow vector; and are the node heating temperature and outlet temperature vector respectively. When the node is a heat source node, the outlet temperature vector is the return water temperature of the node; and Respectively represent the flow temperature at the beginning and end of the pipeline; T a is the ambient temperature; λ is the total heat transfer coefficient per unit length of each pipe; L represents the length of the pipe; and They represent the pipe flow and its temperature before the node i respectively; and represents the pipe flow out of node i and the temperature of node i after the flow is mixed; B is the loop-pipe association matrix; K is the pipe resistance coefficient vector; f is the friction factor, whose size is determined by the Reynolds number R e Determine; D is the pipe diameter; ρ is the density of water; g represents the acceleration due to gravity; Formula (6) defines the energy interaction relationship of the heating network to ensure the energy conservation of the heating network; Formula (7) reflects the temperature loss of the flow in the heating network pipeline; Formula (8) is the physical constraint that needs to be satisfied after the flow is mixed at a node; Formula (9) is the physical constraint that needs to be followed for the flow in the heating network pipeline to maintain circulation; Based on formula (7) In actual operation, the term is far less than 1 and close to 0. The first-order Taylor expansion of the exponential term is performed to obtain the energy loss expression of the heat network pipeline flow (10): Where: is the energy loss caused by the flow passing through the pipeline between nodes i and j; They represent the energy of the flow at the beginning and end of the pipeline respectively; It can be seen from formula (10) that after linearization, the heat network loss will be simplified to a constant term that is only related to the pipeline parameters itself; based on this, the heat network pipeline flow loss is incorporated into the load of each node, thus obtaining the heat network energy balance constraint shown in formula (11): Where: ΔΦ t Is The pipeline loss vector A l is the correlation matrix between the heat network node and the pipeline flow into the node; In actual operation of the heating network, f=64 / R e , while R e =4|m t | / μρπD, μ is the kinematic viscosity of water, then equation (11) can be linearized into equation (12); so far, the energy transfer model of the thermal system is only a linear combination of the pipeline flow rate, realizing the linearization of the energy transfer model of the thermal system: Where: R f is the head loss coefficient vector.

2. The double-layer affine optimization scheduling method for thermoelectric micro-energy grid considering source-load uncertainty according to claim 1 is characterized in that: In the above 2), based on the double-layer affine optimization scheduling method of the thermoelectric micro-energy network with the upper layer being the interval linear programming model and the lower layer being the affine optimization model, a double-layer affine optimization scheduling method of the thermoelectric micro-energy network is proposed according to the different needs of the thermoelectric micro-energy network day-ahead scheduling; the details are as follows: 2.1) Day-ahead optimization objective function While ensuring optimal scheduling and taking into account carbon emission requirements, the objective function constructed is shown in formula (13), which includes minimizing the unit capacity cost, carbon emission penalty cost, equipment operation and maintenance cost, energy abandonment cost, main grid power exchange cost and unit start-up and shutdown cost: in: Where: F fuel is the capacity cost of the CHP unit, F i steam is the capacity cost of the steam CHP unit, F i gas is the capacity cost of the gas-fired CHP unit, and are the electrical output and thermal output of the steam CHP unit, is the electrical output of the gas-fired CHP unit, N s and N g are the total number of steam CHP units and gas CHP units, is a 0-1 variable, indicating the working status of the gas-fired CHP unit; F co2 is the carbon emission penalty cost, γ represents the carbon emission intensity of the CHP unit; F om is the equipment operation and maintenance cost, N d is the total number of devices, F is the equipment output value; s is the energy abandonment cost, and are the values ​​of abandoned solar power, abandoned wind power and abandoned load respectively; F trade is the main network power interaction cost, and Respectively represent the purchase and sale of electricity to the main network; F UC is the unit start-up and shutdown cost, and are 0-1 variables, representing the startup and shutdown actions of the two types of CHP units, j∈{s,g}; c co2 , c sPv 、c sPw 、c sp , They represent the corresponding cost coefficients respectively; Δt is the duration of a single scheduling period; 2.2) Upper level interval linear programming model An upper interval linear programming model is constructed, and interval numbers are used to describe the random fluctuations of renewable energy output and load. An interval linear optimization method is proposed to solve the day-ahead unit combination problem of thermal power micro-energy grid. The details are as follows: When the output and load of renewable energy are modeled as interval numbers according to the predicted fluctuation boundary, the continuous variables including unit output, pipeline flow, node voltage, and energy storage device output will be transformed into corresponding interval forms, so that the unit combination problem of the thermal power micro-energy network becomes an uncertainty interval optimization problem; the interval optimization problem is solved by transforming the original uncertainty optimization model containing interval variables into two deterministic sub-models, namely the upper limit sub-model and the lower limit sub-model, so as to obtain the upper limit of the cost under the worst conditions and the lower limit of the cost under the optimal conditions, thereby solving the interval optimization problem; based on the constructed linear energy transmission model of the thermal power micro-energy network, the constraints in the interval optimization model are all linear constraints; on this basis, considering that the objective function is a monotonic convex function, it can be known that The worst and best conditions that make the objective function take the upper and lower limit values ​​will appear on the fluctuation boundary of the new energy output and load; combined with the highly positive correlation between the objective function value and the unit output, it can be obtained that the worst condition is the maximum unit output, and the best condition is the minimum unit output, that is, the worst condition is that the new energy output changes to the lower limit of the interval and the load changes to the upper limit of the interval, and the best condition is that the new energy output changes to the upper limit of the interval and the load changes to the lower limit of the interval; the worst and best conditions can be used to transform the uncertain interval optimization model into a deterministic upper limit sub-model and a lower limit sub-model. By solving the corresponding sub-models, the uncertain unit combination problem can be solved, and a unit combination interval scheduling plan can be provided for the construction of the lower-level affine optimization model; 2.3) Lower layer affine optimization model Based on the unit combination interval scheduling plan obtained by solving the upper interval linear programming model, the initial optimal operating point of each unit combination plan is further solved when the output of new energy and load fluctuations are not considered. The Taylor expansion mentioned in the linear energy transmission model of the thermal micro-energy network is performed at this operating point, and the linear energy transmission model of the thermal micro-energy network taking into account the loss is constructed. On this basis, the lower-level affine optimization model considering the uncertainty of new energy output and load fluctuations is constructed.

Images